source: MondoRescue/branches/stable/mindi-busybox/archival/libunarchive/decompress_unlzma.c@ 821

Last change on this file since 821 was 821, checked in by Bruno Cornec, 18 years ago

Addition of busybox 1.2.1 as a mindi-busybox new package
This should avoid delivering binary files in mindi not built there (Fedora and Debian are quite serious about that)
File size: 11.7 KB
Line 
1/* vi:set ts=4: */
2/*
3 * Small lzma deflate implementation.
4 * Copyright (C) 2006 Aurelien Jacobs <aurel@gnuage.org>
5 *
6 * Based on LzmaDecode.c from the LZMA SDK 4.22 (http://www.7-zip.org/)
7 * Copyright (C) 1999-2005 Igor Pavlov
8 *
9 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
10 */
11
12#include "libbb.h"
13#include "unarchive.h"
14
15#ifdef CONFIG_FEATURE_LZMA_FAST
16# define speed_inline ATTRIBUTE_ALWAYS_INLINE
17#else
18# define speed_inline
19#endif
20
21
22typedef struct {
23 int fd;
24 uint8_t *ptr;
25 uint8_t *buffer;
26 uint8_t *buffer_end;
27 int buffer_size;
28 uint32_t code;
29 uint32_t range;
30 uint32_t bound;
31} rc_t;
32
33
34#define RC_TOP_BITS 24
35#define RC_MOVE_BITS 5
36#define RC_MODEL_TOTAL_BITS 11
37
38
39/* Called twice: once at startup and once in rc_normalize() */
40static void rc_read(rc_t * rc)
41{
42 rc->buffer_size = read(rc->fd, rc->buffer, rc->buffer_size);
43 if (rc->buffer_size <= 0)
44 bb_error_msg_and_die("unexpected EOF");
45 rc->ptr = rc->buffer;
46 rc->buffer_end = rc->buffer + rc->buffer_size;
47}
48
49/* Called once */
50static void rc_init(rc_t * rc, int fd, int buffer_size)
51{
52 int i;
53
54 rc->fd = fd;
55 rc->buffer = xmalloc(buffer_size);
56 rc->buffer_size = buffer_size;
57 rc->buffer_end = rc->buffer + rc->buffer_size;
58 rc->ptr = rc->buffer_end;
59
60 rc->code = 0;
61 rc->range = 0xFFFFFFFF;
62 for (i = 0; i < 5; i++) {
63 if (rc->ptr >= rc->buffer_end)
64 rc_read(rc);
65 rc->code = (rc->code << 8) | *rc->ptr++;
66 }
67}
68
69/* Called once. TODO: bb_maybe_free() */
70static ATTRIBUTE_ALWAYS_INLINE void rc_free(rc_t * rc)
71{
72 if (ENABLE_FEATURE_CLEAN_UP)
73 free(rc->buffer);
74}
75
76/* Called twice, but one callsite is in speed_inline'd rc_is_bit_0_helper() */
77static void rc_do_normalize(rc_t * rc)
78{
79 if (rc->ptr >= rc->buffer_end)
80 rc_read(rc);
81 rc->range <<= 8;
82 rc->code = (rc->code << 8) | *rc->ptr++;
83}
84static ATTRIBUTE_ALWAYS_INLINE void rc_normalize(rc_t * rc)
85{
86 if (rc->range < (1 << RC_TOP_BITS)) {
87 rc_do_normalize(rc);
88 }
89}
90
91/* Called 9 times */
92/* Why rc_is_bit_0_helper exists?
93 * Because we want to always expose (rc->code < rc->bound) to optimizer
94 */
95static speed_inline uint32_t rc_is_bit_0_helper(rc_t * rc, uint16_t * p)
96{
97 rc_normalize(rc);
98 rc->bound = *p * (rc->range >> RC_MODEL_TOTAL_BITS);
99 return rc->bound;
100}
101static ATTRIBUTE_ALWAYS_INLINE int rc_is_bit_0(rc_t * rc, uint16_t * p)
102{
103 uint32_t t = rc_is_bit_0_helper(rc, p);
104 return rc->code < t;
105}
106
107/* Called ~10 times, but very small, thus inlined */
108static speed_inline void rc_update_bit_0(rc_t * rc, uint16_t * p)
109{
110 rc->range = rc->bound;
111 *p += ((1 << RC_MODEL_TOTAL_BITS) - *p) >> RC_MOVE_BITS;
112}
113static speed_inline void rc_update_bit_1(rc_t * rc, uint16_t * p)
114{
115 rc->range -= rc->bound;
116 rc->code -= rc->bound;
117 *p -= *p >> RC_MOVE_BITS;
118}
119
120/* Called 4 times in unlzma loop */
121static int rc_get_bit(rc_t * rc, uint16_t * p, int *symbol)
122{
123 if (rc_is_bit_0(rc, p)) {
124 rc_update_bit_0(rc, p);
125 *symbol *= 2;
126 return 0;
127 } else {
128 rc_update_bit_1(rc, p);
129 *symbol = *symbol * 2 + 1;
130 return 1;
131 }
132}
133
134/* Called once */
135static ATTRIBUTE_ALWAYS_INLINE int rc_direct_bit(rc_t * rc)
136{
137 rc_normalize(rc);
138 rc->range >>= 1;
139 if (rc->code >= rc->range) {
140 rc->code -= rc->range;
141 return 1;
142 }
143 return 0;
144}
145
146/* Called twice */
147static speed_inline void
148rc_bit_tree_decode(rc_t * rc, uint16_t * p, int num_levels, int *symbol)
149{
150 int i = num_levels;
151
152 *symbol = 1;
153 while (i--)
154 rc_get_bit(rc, p + *symbol, symbol);
155 *symbol -= 1 << num_levels;
156}
157
158
159typedef struct {
160 uint8_t pos;
161 uint32_t dict_size;
162 uint64_t dst_size;
163} __attribute__ ((packed)) lzma_header_t;
164
165
166#define LZMA_BASE_SIZE 1846
167#define LZMA_LIT_SIZE 768
168
169#define LZMA_NUM_POS_BITS_MAX 4
170
171#define LZMA_LEN_NUM_LOW_BITS 3
172#define LZMA_LEN_NUM_MID_BITS 3
173#define LZMA_LEN_NUM_HIGH_BITS 8
174
175#define LZMA_LEN_CHOICE 0
176#define LZMA_LEN_CHOICE_2 (LZMA_LEN_CHOICE + 1)
177#define LZMA_LEN_LOW (LZMA_LEN_CHOICE_2 + 1)
178#define LZMA_LEN_MID (LZMA_LEN_LOW \
179 + (1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_LOW_BITS)))
180#define LZMA_LEN_HIGH (LZMA_LEN_MID \
181 +(1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_MID_BITS)))
182#define LZMA_NUM_LEN_PROBS (LZMA_LEN_HIGH + (1 << LZMA_LEN_NUM_HIGH_BITS))
183
184#define LZMA_NUM_STATES 12
185#define LZMA_NUM_LIT_STATES 7
186
187#define LZMA_START_POS_MODEL_INDEX 4
188#define LZMA_END_POS_MODEL_INDEX 14
189#define LZMA_NUM_FULL_DISTANCES (1 << (LZMA_END_POS_MODEL_INDEX >> 1))
190
191#define LZMA_NUM_POS_SLOT_BITS 6
192#define LZMA_NUM_LEN_TO_POS_STATES 4
193
194#define LZMA_NUM_ALIGN_BITS 4
195
196#define LZMA_MATCH_MIN_LEN 2
197
198#define LZMA_IS_MATCH 0
199#define LZMA_IS_REP (LZMA_IS_MATCH + (LZMA_NUM_STATES <<LZMA_NUM_POS_BITS_MAX))
200#define LZMA_IS_REP_G0 (LZMA_IS_REP + LZMA_NUM_STATES)
201#define LZMA_IS_REP_G1 (LZMA_IS_REP_G0 + LZMA_NUM_STATES)
202#define LZMA_IS_REP_G2 (LZMA_IS_REP_G1 + LZMA_NUM_STATES)
203#define LZMA_IS_REP_0_LONG (LZMA_IS_REP_G2 + LZMA_NUM_STATES)
204#define LZMA_POS_SLOT (LZMA_IS_REP_0_LONG \
205 + (LZMA_NUM_STATES << LZMA_NUM_POS_BITS_MAX))
206#define LZMA_SPEC_POS (LZMA_POS_SLOT \
207 +(LZMA_NUM_LEN_TO_POS_STATES << LZMA_NUM_POS_SLOT_BITS))
208#define LZMA_ALIGN (LZMA_SPEC_POS \
209 + LZMA_NUM_FULL_DISTANCES - LZMA_END_POS_MODEL_INDEX)
210#define LZMA_LEN_CODER (LZMA_ALIGN + (1 << LZMA_NUM_ALIGN_BITS))
211#define LZMA_REP_LEN_CODER (LZMA_LEN_CODER + LZMA_NUM_LEN_PROBS)
212#define LZMA_LITERAL (LZMA_REP_LEN_CODER + LZMA_NUM_LEN_PROBS)
213
214
215int unlzma(int src_fd, int dst_fd)
216{
217 lzma_header_t header;
218 int lc, pb, lp;
219 uint32_t pos_state_mask;
220 uint32_t literal_pos_mask;
221 uint32_t pos;
222 uint16_t *p;
223 uint16_t *prob;
224 uint16_t *prob_lit;
225 int num_bits;
226 int num_probs;
227 rc_t rc;
228 int i, mi;
229 uint8_t *buffer;
230 uint8_t previous_byte = 0;
231 size_t buffer_pos = 0, global_pos = 0;
232 int len = 0;
233 int state = 0;
234 uint32_t rep0 = 1, rep1 = 1, rep2 = 1, rep3 = 1;
235
236 if (read(src_fd, &header, sizeof(header)) != sizeof(header))
237 bb_error_msg_and_die("can't read header");
238
239 if (header.pos >= (9 * 5 * 5))
240 bb_error_msg_and_die("bad header");
241 mi = header.pos / 9;
242 lc = header.pos % 9;
243 pb = mi / 5;
244 lp = mi % 5;
245 pos_state_mask = (1 << pb) - 1;
246 literal_pos_mask = (1 << lp) - 1;
247
248 header.dict_size = SWAP_LE32(header.dict_size);
249 header.dst_size = SWAP_LE64(header.dst_size);
250
251 if (header.dict_size == 0)
252 header.dict_size = 1;
253
254 buffer = xmalloc(MIN(header.dst_size, header.dict_size));
255
256 num_probs = LZMA_BASE_SIZE + (LZMA_LIT_SIZE << (lc + lp));
257 p = xmalloc(num_probs * sizeof(*p));
258 num_probs = LZMA_LITERAL + (LZMA_LIT_SIZE << (lc + lp));
259 for (i = 0; i < num_probs; i++)
260 p[i] = (1 << RC_MODEL_TOTAL_BITS) >> 1;
261
262 rc_init(&rc, src_fd, 0x10000);
263
264 while (global_pos + buffer_pos < header.dst_size) {
265 int pos_state = (buffer_pos + global_pos) & pos_state_mask;
266
267 prob =
268 p + LZMA_IS_MATCH + (state << LZMA_NUM_POS_BITS_MAX) + pos_state;
269 if (rc_is_bit_0(&rc, prob)) {
270 mi = 1;
271 rc_update_bit_0(&rc, prob);
272 prob = (p + LZMA_LITERAL + (LZMA_LIT_SIZE
273 * ((((buffer_pos + global_pos) & literal_pos_mask) << lc)
274 + (previous_byte >> (8 - lc)))));
275
276 if (state >= LZMA_NUM_LIT_STATES) {
277 int match_byte;
278
279 pos = buffer_pos - rep0;
280 while (pos >= header.dict_size)
281 pos += header.dict_size;
282 match_byte = buffer[pos];
283 do {
284 int bit;
285
286 match_byte <<= 1;
287 bit = match_byte & 0x100;
288 prob_lit = prob + 0x100 + bit + mi;
289 if (rc_get_bit(&rc, prob_lit, &mi)) {
290 if (!bit)
291 break;
292 } else {
293 if (bit)
294 break;
295 }
296 } while (mi < 0x100);
297 }
298 while (mi < 0x100) {
299 prob_lit = prob + mi;
300 rc_get_bit(&rc, prob_lit, &mi);
301 }
302 previous_byte = (uint8_t) mi;
303
304 buffer[buffer_pos++] = previous_byte;
305 if (buffer_pos == header.dict_size) {
306 buffer_pos = 0;
307 global_pos += header.dict_size;
308 write(dst_fd, buffer, header.dict_size);
309 }
310 if (state < 4)
311 state = 0;
312 else if (state < 10)
313 state -= 3;
314 else
315 state -= 6;
316 } else {
317 int offset;
318 uint16_t *prob_len;
319
320 rc_update_bit_1(&rc, prob);
321 prob = p + LZMA_IS_REP + state;
322 if (rc_is_bit_0(&rc, prob)) {
323 rc_update_bit_0(&rc, prob);
324 rep3 = rep2;
325 rep2 = rep1;
326 rep1 = rep0;
327 state = state < LZMA_NUM_LIT_STATES ? 0 : 3;
328 prob = p + LZMA_LEN_CODER;
329 } else {
330 rc_update_bit_1(&rc, prob);
331 prob = p + LZMA_IS_REP_G0 + state;
332 if (rc_is_bit_0(&rc, prob)) {
333 rc_update_bit_0(&rc, prob);
334 prob = (p + LZMA_IS_REP_0_LONG
335 + (state << LZMA_NUM_POS_BITS_MAX) + pos_state);
336 if (rc_is_bit_0(&rc, prob)) {
337 rc_update_bit_0(&rc, prob);
338
339 state = state < LZMA_NUM_LIT_STATES ? 9 : 11;
340 pos = buffer_pos - rep0;
341 while (pos >= header.dict_size)
342 pos += header.dict_size;
343 previous_byte = buffer[pos];
344 buffer[buffer_pos++] = previous_byte;
345 if (buffer_pos == header.dict_size) {
346 buffer_pos = 0;
347 global_pos += header.dict_size;
348 write(dst_fd, buffer, header.dict_size);
349 }
350 continue;
351 } else {
352 rc_update_bit_1(&rc, prob);
353 }
354 } else {
355 uint32_t distance;
356
357 rc_update_bit_1(&rc, prob);
358 prob = p + LZMA_IS_REP_G1 + state;
359 if (rc_is_bit_0(&rc, prob)) {
360 rc_update_bit_0(&rc, prob);
361 distance = rep1;
362 } else {
363 rc_update_bit_1(&rc, prob);
364 prob = p + LZMA_IS_REP_G2 + state;
365 if (rc_is_bit_0(&rc, prob)) {
366 rc_update_bit_0(&rc, prob);
367 distance = rep2;
368 } else {
369 rc_update_bit_1(&rc, prob);
370 distance = rep3;
371 rep3 = rep2;
372 }
373 rep2 = rep1;
374 }
375 rep1 = rep0;
376 rep0 = distance;
377 }
378 state = state < LZMA_NUM_LIT_STATES ? 8 : 11;
379 prob = p + LZMA_REP_LEN_CODER;
380 }
381
382 prob_len = prob + LZMA_LEN_CHOICE;
383 if (rc_is_bit_0(&rc, prob_len)) {
384 rc_update_bit_0(&rc, prob_len);
385 prob_len = (prob + LZMA_LEN_LOW
386 + (pos_state << LZMA_LEN_NUM_LOW_BITS));
387 offset = 0;
388 num_bits = LZMA_LEN_NUM_LOW_BITS;
389 } else {
390 rc_update_bit_1(&rc, prob_len);
391 prob_len = prob + LZMA_LEN_CHOICE_2;
392 if (rc_is_bit_0(&rc, prob_len)) {
393 rc_update_bit_0(&rc, prob_len);
394 prob_len = (prob + LZMA_LEN_MID
395 + (pos_state << LZMA_LEN_NUM_MID_BITS));
396 offset = 1 << LZMA_LEN_NUM_LOW_BITS;
397 num_bits = LZMA_LEN_NUM_MID_BITS;
398 } else {
399 rc_update_bit_1(&rc, prob_len);
400 prob_len = prob + LZMA_LEN_HIGH;
401 offset = ((1 << LZMA_LEN_NUM_LOW_BITS)
402 + (1 << LZMA_LEN_NUM_MID_BITS));
403 num_bits = LZMA_LEN_NUM_HIGH_BITS;
404 }
405 }
406 rc_bit_tree_decode(&rc, prob_len, num_bits, &len);
407 len += offset;
408
409 if (state < 4) {
410 int pos_slot;
411
412 state += LZMA_NUM_LIT_STATES;
413 prob =
414 p + LZMA_POS_SLOT +
415 ((len <
416 LZMA_NUM_LEN_TO_POS_STATES ? len :
417 LZMA_NUM_LEN_TO_POS_STATES - 1)
418 << LZMA_NUM_POS_SLOT_BITS);
419 rc_bit_tree_decode(&rc, prob, LZMA_NUM_POS_SLOT_BITS,
420 &pos_slot);
421 if (pos_slot >= LZMA_START_POS_MODEL_INDEX) {
422 num_bits = (pos_slot >> 1) - 1;
423 rep0 = 2 | (pos_slot & 1);
424 if (pos_slot < LZMA_END_POS_MODEL_INDEX) {
425 rep0 <<= num_bits;
426 prob = p + LZMA_SPEC_POS + rep0 - pos_slot - 1;
427 } else {
428 num_bits -= LZMA_NUM_ALIGN_BITS;
429 while (num_bits--)
430 rep0 = (rep0 << 1) | rc_direct_bit(&rc);
431 prob = p + LZMA_ALIGN;
432 rep0 <<= LZMA_NUM_ALIGN_BITS;
433 num_bits = LZMA_NUM_ALIGN_BITS;
434 }
435 i = 1;
436 mi = 1;
437 while (num_bits--) {
438 if (rc_get_bit(&rc, prob + mi, &mi))
439 rep0 |= i;
440 i <<= 1;
441 }
442 } else
443 rep0 = pos_slot;
444 if (++rep0 == 0)
445 break;
446 }
447
448 len += LZMA_MATCH_MIN_LEN;
449
450 do {
451 pos = buffer_pos - rep0;
452 while (pos >= header.dict_size)
453 pos += header.dict_size;
454 previous_byte = buffer[pos];
455 buffer[buffer_pos++] = previous_byte;
456 if (buffer_pos == header.dict_size) {
457 buffer_pos = 0;
458 global_pos += header.dict_size;
459 write(dst_fd, buffer, header.dict_size);
460 }
461 len--;
462 } while (len != 0 && buffer_pos < header.dst_size);
463 }
464 }
465
466 write(dst_fd, buffer, buffer_pos);
467 rc_free(&rc);
468 return 0;
469}
470
471/* vi:set ts=4: */
Note: See TracBrowser for help on using the repository browser.